scholarly journals DEEP LEARNING TO PREDICT DEGREE OF INTERSTITIAL FIBROSIS AND TUBULAR ATROPHY FROM KIDNEY ULTRASOUND IMAGES - AN ARTIFICIAL INTELLIGENCE APPROACH

Author(s):  
Ambarish M Athavale ◽  
Peter D Hart ◽  
Matthew Itteera ◽  
Tushar Patel ◽  
David J Cymbaluk ◽  
...  

Background: Interstitial fibrosis and tubular atrophy (IFTA) is a strong predictor of decline in kidney function. Non-invasive test to assess IFTA is not available. Methods: We trained, validated and tested a deep learning (DL) system to classify IFTA grade from 6,135 ultrasound images obtained from 352 patients who underwent kidney biopsy. Of 6,135 ultrasound images, 5,523 were used for training (n = 5,122) and validation (n = 401) and 612 to test the accuracy of the DL system. IFTA grade scored by nephropathologist on trichrome stained kidney biopsy slide was used as reference standard. Results: There were 159 patients (2,701 ultrasound images), 74 patients (1,239 ultrasound images), 41 patients (701 ultrasound images) and 78 patients (1,494 ultrasound images) with IFTA grades 1, 2, 3 and 4, respectively. The deep-learning classification system used masked images based on a 91% accurate kidney segmentation routine. The performance matrices for the deep learning classifier algorithm in the validation set showed excellent precision (90%), recall (76%), accuracy (84%) and F1-score (80%). In the independent test set also, performance matrices showed excellent precision (90%), recall (80%), accuracy (87%) and F1-score of (84%). Accuracy was highest for IFTA grade 1 (98%) and IFTA grade 4 (82%). Conclusion: A DL system can accurately predict IFTA from kidney ultrasound image.

2016 ◽  
Vol 44 (4) ◽  
pp. 286-288 ◽  
Author(s):  
Musab S. Hommos ◽  
Andrew D. Rule

Severity of interstitial fibrosis and tubular atrophy (IFTA) seen on kidney biopsy has been used by nephrologist as prognostic marker of kidney disease. While IFTA is a very strong predictor of kidney failure, some patients with severe fibrosis will still not progress to kidney failure within 5 years. Treatment of underlying kidney disease in select patients with severe IFTA should be considered when risks of treatment are reasonable despite the low potential for benefit.


2021 ◽  
pp. ASN.2020081181 ◽  
Author(s):  
Aishwarya Ravindran ◽  
Marta Casal Moura ◽  
Fernando C. Fervenza ◽  
Samih H. Nasr ◽  
Mariam P. Alexander ◽  
...  

BackgroundIn patients with secondary (autoimmune) membranous nephropathy, two novel proteins, Exostosin 1 and Exostosin 2 (EXT1/EXT2), are potential disease antigens, biomarkers, or both. In this study, we validate the EXT1/EXT2 findings in a large cohort of membranous lupus nephritis.MethodsWe conducted a retrospective cohort study of patients with membranous lupus nephritis, and performed immunohistochemistry studies on the kidney biopsy specimens against EXT1 and EXT2. Clinicopathologic features and outcomes of EXT1/EXT2-positive versus EXT1/EXT2-negative patients were compared.ResultsOur study cohort included 374 biopsy-proven membranous lupus nephritis cases, of which 122 (32.6%) were EXT1/EXT2-positive and 252 (67.4%) were EXT1/EXT2-negative. EXT1/EXT2-positive patients were significantly younger (P=0.01), had significantly lower serum creatinine levels (P=0.02), were significantly more likely to present with proteinuria ≥3.5 g/24 h (P=0.009), and had significantly less chronicity features (glomerulosclerosis, P=0.001 or interstitial fibrosis and tubular atrophy, P<0.001) on kidney biopsy. Clinical follow-up data were available for 160 patients, of which 64 (40%) biopsy results were EXT1/EXT2-positive and 96 (60%) were EXT1/EXT2-negative. The proportion of patients with class 3/4 lupus nephritis coexisting with membranous lupus nephritis was not different between the EXT1/EXT2-positive and EXT1/EXT2-negative groups (25.0% versus 32.3%; P=0.32). The patients who were EXT1/EXT2-negative evolved to ESKD faster and more frequently compared with EXT1/EXT2-positive patients (18.8% versus 3.1%; P=0.003).ConclusionsThe prevalence of EXT1/EXT2 positivity was 32.6% in our cohort of membranous lupus nephritis. Compared with EXT1/EXT2-negative membranous lupus nephritis, EXT1/EXT2-positive disease appears to represent a subgroup with favorable kidney biopsy findings with respect to chronicity indices. Cases of membranous lupus nephritis that are EXT1/EXT2-negative are more likely to progress to ESKD compared with those that are EXT1/EXT2-positive.


2021 ◽  
pp. 29-42
Author(s):  
admin admin ◽  
◽  
◽  
Adnan Mohsin Abdulazeez

With the development of technology and smart devices in the medical field, the computer system has become an essential part of this development to learn devices in the medical field. One of the learning methods is deep learning (DL), which is a branch of machine learning (ML). The deep learning approach has been used in this field because it is one of the modern methods of obtaining accurate results through its algorithms, and among these algorithms that are used in this field are convolutional neural networks (CNN) and recurrent neural networks (RNN). In this paper we reviewed what have researchers have done in their researches to solve fetal problems, then summarize and carefully discuss the applications in different tasks identified for segmentation and classification of ultrasound images. Finally, this study discussed the potential challenges and directions for applying deep learning in ultrasound image analysis.


2021 ◽  
Vol 36 (Supplement_1) ◽  
Author(s):  
Oana Ion ◽  
Valentina-Georgiana Fratila ◽  
Andreea Gabriella Andronesi ◽  
Camelia Achim ◽  
Elena Georgia Micu ◽  
...  

Abstract Background and Aims Distal renal tubular acidosis (DRTA) is characterized by hyperchloremic metabolic acidosis, with normal anion gap and with the inability to acidify the urine to a pH lower than 5.3. DRTA can evolve without symptoms and systemic acidosis, this form being defined as incomplete DRTA, that necessitates the use of a urinary acidification test like the Furosemide and Fludrocortisone test for establishing the diagnosis. There are several cases of type 1 DRTA reported in patients with autoimmune diseases, especially Sjögren syndrome and systemic lupus erythematosus (SLE), most of them being diagnosed due to severe symptoms caused by the associated hypokalemia. The prevalence of the DRTA in patients with autoimmune diseases is unknown, especially if we refer to the incomplete form. Method We conducted an observational prospective study in a selected cohort of 21 patients diagnosed with autoimmune diseases, who presented for evaluation in our clinic during December, 2020 and January, 2021. We included patients diagnosed with SLE, Sjögren syndrome, ANCA vasculitis, cryoglobulinemic vasculitis, who were submitted to Furosemide/Fludrocortisone test in our nephrology department. The urinary pH was evaluated hourly for 6 hours after the test began.  The test was considered positive if the urinary pH did not decrease &lt; 5.3. Results The study included 21 patients (16 females, 5 males, mean age 41.52 ± 17.58 years), diagnosed with SLE (13 patients, mean age 30.23 ± 10.34 years, eGFR 81.61±20.39 ml/min/1.73 m2), pANCA vasculitis (6 patients, mean age 60.83 ± 6.14, eGFR 40 ± 12.64 ml/min/1.73 m2), Sjogren syndrome (one 44 year-old patient, eGFR 39 ml/min/1.73 m2) and cryoglobulinemic vasculitis (one 69 year-old patient, eGFR 31 ml/min/1.73 m2). The test was positive for 4 patients out of 21 (3 females, one male; one with SLE, one with pANCA vasculitis, one with Sjogren syndrome and one with cryoglobulinemic vasculitis). Although 2 patients developed hypokalemia defined as a level of serum potassium lower than 3.5 mmol/l after the test and 1 patient augmented previous hypokalemia, there was not a significant change in kalemia (3.93 ± 0.32 mmol/l before the test vs 3.95 ± 0.49 mmol/l after the test, p=0.835). Although none of the patients developed metabolic alkalosis after the test, there was a significant increase in the level of serum bicarbonate (26.6 (2.2) mmol/l before the test vs 28.2 (2.7) mmol/l after the test, p=0.005) and also in the level of serum pH (7.36 ± 0.04 before the test vs 7.38 ± 0.04 after the test, p=0.018). None of the patients reported digestive or allergic side effects. It was interesting that the patients with vasculitis responded with delay to the treatment and urinary acidification under the pH of 5.3 occurred after a mean period of 3.2 hours in comparison to 1.5 hours in patients with SLE (p=0.014). Regarding the histological data, both the patients with vasculitis were elders, with an altered kidney function (both with a eGFR of 31 ml/min/1.73 m2) and severe tubular atrophy and interstitial fibrosis on kidney biopsy. The female patient with cryoglobulinemic vasculitis also had positive titers for antinuclear antibody, anti Ro-antibodies and anti-La antibodies. The patient with Sjögren syndrome was diagnosed with nephrocalcinosis and the kidney biopsy was not effectuated. The youngest patient with a positive test had preserved renal function, without tubular or interstitial lesions on kidney biopsy, but with a pattern of membranous lupus nephritis and with intense immunological activity (ANA, anti ds-DNA antibodies, anti RNP and Sm antibodies, antiphospholipid syndrome). Conclusion Incomplete DRTA was found in 4 out of 21 patients with autoimmune diseases, one with Sjogren syndrome and nephrocalcinosis, two with pANCA and cryoglobulinemic vasculitis with decreased eGFR and severe tubular atrophy and interstitial fibrosis and one young female with SLE.


Lupus ◽  
2020 ◽  
Vol 30 (1) ◽  
pp. 25-34
Author(s):  
Enrique Morales ◽  
Hernando Trujillo ◽  
Teresa Bada ◽  
Marina Alonso ◽  
Eduardo Gutiérrez ◽  
...  

Introduction Recent studies with protocol biopsies have shown a mismatch between clinical and histological remission in lupus nephritis (LN). We aimed to evaluate histological changes in repeat kidney biopsies by clinical indication in patients with LN. Methods We analyzed 107 patients with LN in which a kidney biopsy was performed between 2008 and 2018. Of those, we included 26 (24.2%) who had ≥2 kidney biopsies. Classification was done according to the International Society of Nephrology/Renal Pathology Society. Results Mean time between biopsies was 71.5 ± 10.7 months. 73.1% of patients presented a change of class at repeat biopsy; 38.4% to a higher class and 34.6% to a lower class. A significant increase in glomerulosclerosis (% GS) (3.8% vs 18.7%, p = 0.006), interstitial fibrosis (3.8% vs 26.9%, p = 0.021), tubular atrophy (15.4% vs 57.7%, p = 0.001) and chronicity index (CI) (1 vs 3, p < 0.001) was observed at repeat biopsy. Subjects who developed chronic kidney disease progression had a lower rate of complete remission at 12 months (0% vs 37.5%, p = 0.02), higher % GS at first biopsy (7.9% vs 1.2%, p = 0.02) and higher CI (4 vs 2, p = 0.006), tubular atrophy (90% vs 37.6%, p = 0.008), interstitial fibrosis (50% vs 12.5%, p = 0.036) and vascular lesions (60% vs 18.8%, p = 0.031) at second biopsy. Conclusions Our major finding was that patients with LN showed a significant increase in % GS, interstitial fibrosis, tubular atrophy and vascular lesions in repeat biopsies performed by clinical indication. This suggest that a second kidney biopsy may provide valuable and useful information regarding kidney disease progression.


2021 ◽  
Vol 2021 ◽  
pp. 1-12
Author(s):  
Zhemin Zhuang ◽  
Zengbiao Yang ◽  
Shuxin Zhuang ◽  
Alex Noel Joseph Raj ◽  
Ye Yuan ◽  
...  

Breast ultrasound examination is a routine, fast, and safe method for clinical diagnosis of breast tumors. In this paper, a classification method based on multi-features and support vector machines was proposed for breast tumor diagnosis. Multi-features are composed of characteristic features and deep learning features of breast tumor images. Initially, an improved level set algorithm was used to segment the lesion in breast ultrasound images, which provided an accurate calculation of characteristic features, such as orientation, edge indistinctness, characteristics of posterior shadowing region, and shape complexity. Simultaneously, we used transfer learning to construct a pretrained model as a feature extractor to extract the deep learning features of breast ultrasound images. Finally, the multi-features were fused and fed to support vector machine for the further classification of breast ultrasound images. The proposed model, when tested on unknown samples, provided a classification accuracy of 92.5% for cancerous and noncancerous tumors.


2021 ◽  
Vol 2021 ◽  
pp. 1-8
Author(s):  
Ji Li ◽  
Dan Liu ◽  
Xiaofeng Qing ◽  
Lanlan Yu ◽  
Huizhen Xiang

This study was aimed to enhance and detect the characteristics of three-dimensional transvaginal ultrasound images based on the partial differential algorithm and HSegNet algorithm under deep learning. Thereby, the effect of quantitative parameter values of optimized three-dimensional ultrasound image was analyzed on the diagnosis and evaluation of intrauterine adhesions. Specifically, 75 patients with suspected intrauterine adhesion in hospital who underwent the hysteroscopic diagnosis were selected as the research subjects. The three-dimensional transvaginal ultrasound image was enhanced and optimized by the partial differential equation algorithm and processed by the deep learning algorithm. Subsequently, three-dimensional transvaginal ultrasound examinations were performed on the study subjects that met the standards. The March classification method was used to classify the patients with intrauterine adhesion. Finally, the results by the three-dimensional transvaginal ultrasound were compared with the diagnosis results in hysteroscope surgery. The results showed that the HSegNet algorithm model realized the automatic labeling of intrauterine adhesion in the transvaginal ultrasound image and the final accuracy coefficient was 97.3%. It suggested that the three-dimensional transvaginal ultrasound diagnosis based on deep learning was efficient and accurate. The accuracy of the three-dimensional transvaginal ultrasound was 97.14%, the sensitivity was 96.6%, and the specificity was 72%. In conclusion, the three-dimensional transvaginal examination can effectively improve the diagnostic efficiency of intrauterine adhesion, providing theoretical support for the subsequent diagnosis and grading of intrauterine adhesion.


Kidney360 ◽  
2021 ◽  
pp. 10.34067/KID.0003662021
Author(s):  
Chanon Chantaduly ◽  
Hayden R. Troutt ◽  
Karla A. Perez Reyes ◽  
Jonathan E. Zuckerman ◽  
Peter D. Chang ◽  
...  

Background: The goal of the Artificial Intelligence in Renal Scarring (AIRS) study is to develop machine learning tools for non-invasive quantification of kidney fibrosis from imaging scans. Methods: We conducted a retrospective analysis of patients who had one or more abdominal computed tomography (CT) scans within 6 months of a kidney biopsy. The final cohort encompassed 152 CT scans from 92 patients which included images of 300 native kidneys and 76 transplant kidneys. Two different convolutional neural networks (slice-level and voxel-level classifiers) were tested to differentiate severe vs mild/moderate kidney fibrosis (≥50% vs <50%). Interstitial fibrosis and tubular atrophy scores from kidney biopsy reports were used as ground-truth. Results: The two machine learning models demonstrated similar positive predictive value (0.886 vs 0.935) and accuracy (0.831 vs 0.879). Conclusions: In summary, machine learning algorithms are a promising non-invasive diagnostic tool to quantify kidney fibrosis from CT scans. The clinical utility of these prediction tools, in terms of avoiding renal biopsy and associated bleeding risks in patients with severe fibrosis, remains to be validated in prospective clinical trials.


2021 ◽  
Vol 10 (18) ◽  
pp. 4191
Author(s):  
Yura Chae ◽  
Hye Eun Yoon ◽  
Yoon Kyung Chang ◽  
Young Soo Kim ◽  
Hyung Wook Kim ◽  
...  

Immunoglobulin M nephropathy (IgMN) is an idiopathic glomerulonephritis characterized by diffuse deposits of IgM in the glomerular mesangium. However, its renal prognosis remains unknown. We compared renal outcomes of IgMN patients with those of patients with minimal change disease (MCD), focal segmental glomerulosclerosis (FSGS), or mesangial proliferative glomerulonephritis (MsPGN) from a prospective observational cohort, with 1791 patients undergoing native kidney biopsy in eight hospitals affiliated with The Catholic University of Korea between December 2014 and October 2020. IgMN had more mesangial proliferation and matrix expansion than MsPGN and more tubular atrophy and interstitial fibrosis than MCD. IgMN patients had decreased eGFR than MCD patients in the earlier follow-up. However, there was no significant difference in urine protein or eGFR among all patients at the last follow-up. When IgMN was divided into three subtypes, patients with FSGS-like IgMN tended to have lower eGFR than those with MCD-like or MsPGN-like IgMN but higher proteinuria than MsPGN-like IgMN without showing a significant difference. The presence of hypertension at the time of kidney biopsy predicted ≥20% decline of eGFR over two years in IgMN patients. Our data indicate that IgMN would have a clinical course and renal prognosis similar to MCD, FSGS, and MsPGN


2021 ◽  
Vol 8 ◽  
Author(s):  
Shuo-Ming Ou ◽  
Ming-Tsun Tsai ◽  
Huan-Yuan Chen ◽  
Fu-An Li ◽  
Wei-Cheng Tseng ◽  
...  

Background: Galectin-3 (Gal-3) is a multifunctional glycan-binding protein shown to be linked to chronic inflammation and fibrogenesis. Plasma Gal-3 is associated with proteinuria and renal dysfunction, but its role has never been confirmed with kidney biopsy results. In our study, we aimed to explore the expression of Gal-3 in biopsy-proven patients, and we tested the hypothesis that chronic kidney disease (CKD) leads to upregulation of plasma Gal-3 expression in corresponding biopsy findings and RNA sequencing analysis.Method: In 249 patients (male/female: 155/94, age: 57.2 ± 16.3 years) who underwent kidney biopsy, plasma levels of Gal-3 were measured to estimate the association of renal fibrosis. Relationships between plasma Gal-3 levels, estimated glomerular filtration rate (eGFR) and renal histology findings were also assessed. We further examined the gene expression of Gal-3 in RNA-sequencing analysis in biopsy-proven patients.Results: Compared to patients without CKD, CKD patients had higher levels of plasma Gal-3 (1,016.3 ± 628.1 pg/mL vs. 811.6 ± 369.6 pg/ml; P = 0.010). Plasma Gal-3 was inversely correlated with eGFR (P = 0.005) but not with proteinuria. Higher Gal-3 levels were associated with interstitial fibrosis, tubular atrophy and vascular intimal fibrosis. RNA-sequencing analysis showed the upregulation of Gal-3 in fibrotic kidney biopsy samples, and the differentially expressed genes were mainly enhanced in immune cell activation and the regulation of cell-cell adhesion.Conclusions: Plasma Gal-3 levels are inverse correlated with eGFR but positively correlated with renal fibrosis, which may be involved in the immune response and associated pathways. These findings support the role of Gal-3 as a predictive marker of renal fibrosis.


Sign in / Sign up

Export Citation Format

Share Document